5:35 a.m.
To add to this.
I think we are trying to put too much into the components, defeating the
whole purpose of the componentization that I was advocating for in the
2014 f2f.
The components need to do minimum things and do them well. They should
go to maintenance relatively early, after that all the specifics need to
be in Hawkular.
One example, today we have calculation of 95 percentile within Metrics
[1], IMO this should be in hawkular (and Metrics need to provide a way
to add that functionality which I think was one thing discussed at f2f).
If I need 97 or 98 or 99 or 99.99 percentile, I should be able to do
that without another release of metrics.
Hawkular (the kettle) should be in charge of "deciding" who (which
components) receive what information based on Hawkular configuration.
[1] I understand that this was a shortcut. No finger pointing
Thomas
On 05/13/2015 12:25 PM, Thomas Heute wrote:
+1
On 05/13/2015 11:53 AM, Michael Burman wrote:
> Hi,
>
> I disagree with the fact where the actual bus-integration component
> should reside. The components themselves should have the capabilities
> of allowing another component to connect to their internal "feeds" but
> the actual component -> bus connector should be in the
> Kettle-repository. Why? The bus-connector is highly dependent on the
> actual bus implementation. If we would change the bus in any way in
> terms of implementation or API, it would require changes to every
> Hawkular component (thus, they wouldn't be decoupled by design)
> instead of only changing the implementation at the bus-component.
>
> - Micke
>
> ----- Original Message -----
> From: "Gary Brown" <gbrown(a)redhat.com>
> To: "Discussions around Hawkular development"
> <hawkular-dev(a)lists.jboss.org>
> Sent: Wednesday, May 13, 2015 11:02:31 AM
> Subject: [Hawkular-dev] The components, glue and kettle
>
> Hi
>
> On yesterday's watercooler discussion the main topic was about how to
> package the individual components, with the "glue" code, within kettle.
>
> There seems to be a general idea to move all bus integration code into
> Hawkular/Kettle repo. Although I agree that the kettle is the right
> place to bring together the components and glue, not sure whether the
> actual code for the glue should reside in that single repo.
>
> My preference would be for all component related code, including that
> component's integration with the bus, to be located in that component
> repo - that way there is a clear owner of the code, and any changes to
> the core APIs are locally dealt across all integration points that may
> be supported.
>
> Then the kettle repo can be responsible for selecting the relevant
> artifacts to build what it requires.
>
> In terms of how the 'core' artifacts from each component should be
> enhanced with the glue - may be the simplest way would be as one
> person suggested - use an overlay to build upon the 'core' war
> artifact to add the glue artifacts (which in general will probably
> just be bus integration). So each component only needs to produce a
> single 'core' war, but also manages the integration artifacts locally.
>
> I think I would prefer this approach over a single ear, as it still
> retains the individual component boundaries but enhances them with
> whatever they need to communicate in the kettle.
>
> It would be even more ideal if jboss modules allowed exploded wars, so
> that we didn't need to use overlays, but I don't believe this is the
> case.
>
> Regards
> Gary
>
> _______________________________________________
> hawkular-dev mailing list
> hawkular-dev(a)lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hawkular-dev
> _______________________________________________
> hawkular-dev mailing list
> hawkular-dev(a)lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hawkular-dev
>